By automatically, I would hope you mean a kind of "wait_for()" method, rather than doing it unseen in the background. At first thought it seems the latter would be terribly inefficient, especially since it is only necessary in relatively rare cases.
No i meant automatically, all the time.
What it would mean is that whenever Watir failed to find an object, it would sleep for a fraction of a second and then try again if the timeout interval hadn't yet been exceeded. If the timeout were set to 0, it should behave exactly as it does today, so you wouldn't need to replace code.
In my spike, the core test suite, which normally runs in 120 seconds, ran in 142 with the timeout set to 0. And it represents a worst case, since it includes lots of negative tests -- a lot more than any regular test suite or scaper application would include. And in any case, i can't explain why it took an extra 22 seconds -- so that is a bug i'd have to track down before i would commit the code.
Time and efficiency are definitely a concern in our screen scrapes. I was already considering overriding existing delays in the code. If this loop were done every time, we would likely be in the position of having to go into the Watir code to take it back out, in order to regain what efficiency we could.
Please let us know where you think Watir is inefficient. Like i said the other day, i'm always tweaking things to make my tests run faster. If you have suggestions, i'm all ears.
Bret
_______________________________________________ Wtr-general mailing list [email protected] http://rubyforge.org/mailman/listinfo/wtr-general
